A Comparative Dataset of Cloud AI Services for Financial Systems: Provider Capabilities, Cost Structures, and Governance Evidence
Main article
Abstract
Choosing a public cloud provider for artificial intelligence workloads is now one of the most consequential infrastructure decisions facing financial institutions, yet the decision remains poorly supported by structured comparative evidence. This study constructs a comparative dataset covering the three dominant providers — Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP) — and benchmarks them on seven dimensions: managed AI service breadth, foundation model and machine-learning catalogue, MLOps tooling depth, cost transparency, operational resilience track record, security and compliance posture, and governance support for regulated workloads. The coded matrix is analysed using descriptive comparison, principal component analysis, k-means clustering, and scenario-weighted suitability scoring. Three financial-services use cases are examined: real-time fraud detection, batch credit risk modelling, and graph-based anti-money laundering. Results indicate that no single provider dominates across all scenarios. AWS leads on ecosystem breadth and on real-time, latency-sensitive inference; Azure leads on governance maturity and regulated-workload alignment with the European Union Artificial Intelligence Act and the Digital Operational Resilience Act; and GCP leads on model-transparency tooling and on per-operation cost for well-defined batch workloads. The paper argues that provider selection should be treated as a scenario-dependent governance decision rather than a global ranking, and offers a transparent matching framework that financial institutions and supervisors can adopt or adapt. The dataset, rubric, and scoring weights are documented at a level of granularity that supports replication and extension to additional providers, workloads, or regulatory regimes.
